layout | title | permalink |
---|---|---|
page |
Machine Learning and Statistics |
/ml-and-stats-overview/ |
Being a somewhat structured person, it's important for me to see how topics relate to each other. With machine learning and statistics being such expansive topics, I struggled (and still do) to understand how all the different methods related to each other. This page is my attempt to structure the various topics/methods of machine learning and statistics. My ultimate goal will be to have a post on each of these topics.
- There are 2 major branches of statistics:
- Description statistics: Summarizing and presenting key characteristics about some data.
- Inferential statistics: Inferring characteristics about a population from a small sample of it. In this branch, you typically start with a hypothesis and then test if your sample follows the hypothesis.
- Probability and statistics deal with questions involving populations and samples, but do so in an "inverse manner" to one another.
- In a probability problem, properties of a population under study are known (e.g. specified distribution of a population), and questions regarding a sample taken from the population are posed and answered.
- In a statistics problem, characteristics of the sample are known and properties of the population are inferred.
- We study probability first before statistics because we need to understand the uncertainity associated with taking a sample from a population. Then we are start to understand what a particular sample can tell us about a population.
- Descriptional Statistics
- Correlation analysis
- Parametric
- Non-Parametric (Kendall and Spearman)
- Correlation analysis
- [Random Variables]({% post_url 2016-02-26-random-variables %})
- Probabilities
- [Joint, Marginal, and Conditional Probabilites]({% post_url 2016-03-20-basic-prob %})
- [Bayes' Rule]({% post_url 2016-04-21-bayes-rule %})
- [Probability Distributions]({% post_url 2016-03-17-prob-distr %})
- Continuous
- Gaussian (Normal) Distribution
- Dirichlet Distribution
- Exponential Distribution
- Chi-Square Distribution
- Weibull Distribution
- Exponential Distribution
- Beta Distribution
- Discrete
- Bernoulli Distribution
- Binomial Distribution (Sum of n independent Bernoulli trails)
- Multinomial Distribution
- Multivariate Hypergeometric Distribution
- If sampling is done without replacement.
- Poission Distribution
- Negative Binomial Distribution
- Beta-binomial distribution
- Continuous
- Hypothesis Testing
- Other
- [Confidence Intervals]({% post_url 2015-08-25-how-to-interpret-a-CI %})
- Power Analysis
- Survival Analysis
- [The Basics of Survival Analysis]({% post_url 2016-05-12-survival-analysis %})
- Kaplan-Meier Curves and the Log-rank Test
- Cox Regression
- Survival Analysis Study Design Considerations
- Kinds of Statistics: Good introductory article on what the two main branches of statistics are.
- Introduction to Statistical Thought
- [Bayesian Inference 101]({% post_url 2017-03-08-how-to-bayesian-infer-101 %})
- Classical Frequentist vs. Bayesian
- Markov Chain Monte Carlo (MCMC) processes
- Metropolis algorithm
- Gibbs Sampling (Special case of Metropolis)
- BUGS, JAGS, STAN
- Bayesian Statistics explained to Beginners in Simple English
- Doing Bayesian Data Analysis - A Tutorial with R, JAGS, and Stan

- Supervised Learning
- Regression (predict continuous values)
- Linear Regression
- Artificial Neural Networks (ANN)
- Can also be used for classification.
- Classification - Predict discrete (categorical) values (i.e. class a data point belongs to)
- Logistic Regression
- Linear Discriminant Analysis (LDA)
- Naive Bayes Classifier
- Support Vector Machines (SVM)
- Random Forest
- ANN
- Can also be used for regression.
- Regression (predict continuous values)
- Unsupervised Learning
- Cluster Analysis
- Principle Component Analysis (PCA)
- t-SNE
- Hierarchical clustering
- DBSCAN
- K-means
- [Mixture Models]({% post_url 2015-10-13-mixture-model %})
- Topic Modeling
- Latent Dirichlet Allocation (LDA)
- Non-negative Matrix Factorization (NMF)
- Cluster Analysis
- Feature Selection
- Other
- Bias-Variance Tradeoff
- Receiver operating characteristic (ROC) curve
- Mathews correlation coefficient (MCC)